wav2vec2-large-xls-r-300m-latvian / trainer_state.json
infinitejoy's picture
End of training
19aada7
raw
history blame
21.2 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 100.0,
"global_step": 15600,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.64,
"learning_rate": 3.3949999999999997e-06,
"loss": 18.8158,
"step": 100
},
{
"epoch": 1.28,
"learning_rate": 6.895e-06,
"loss": 9.3721,
"step": 200
},
{
"epoch": 1.92,
"learning_rate": 1.0394999999999998e-05,
"loss": 5.1453,
"step": 300
},
{
"epoch": 2.56,
"learning_rate": 1.3895e-05,
"loss": 4.2355,
"step": 400
},
{
"epoch": 3.21,
"learning_rate": 1.7395e-05,
"loss": 3.7107,
"step": 500
},
{
"epoch": 3.85,
"learning_rate": 2.0894999999999996e-05,
"loss": 3.3647,
"step": 600
},
{
"epoch": 4.49,
"learning_rate": 2.4394999999999996e-05,
"loss": 3.2191,
"step": 700
},
{
"epoch": 5.13,
"learning_rate": 2.7895e-05,
"loss": 3.1339,
"step": 800
},
{
"epoch": 5.77,
"learning_rate": 3.1395e-05,
"loss": 3.0484,
"step": 900
},
{
"epoch": 6.41,
"learning_rate": 3.4895e-05,
"loss": 3.0086,
"step": 1000
},
{
"epoch": 7.05,
"learning_rate": 3.8394999999999994e-05,
"loss": 2.9794,
"step": 1100
},
{
"epoch": 7.69,
"learning_rate": 4.1895e-05,
"loss": 2.9544,
"step": 1200
},
{
"epoch": 8.33,
"learning_rate": 4.5394999999999995e-05,
"loss": 2.901,
"step": 1300
},
{
"epoch": 8.97,
"learning_rate": 4.8895e-05,
"loss": 2.6256,
"step": 1400
},
{
"epoch": 9.62,
"learning_rate": 5.2395e-05,
"loss": 2.0673,
"step": 1500
},
{
"epoch": 10.26,
"learning_rate": 5.589499999999999e-05,
"loss": 1.763,
"step": 1600
},
{
"epoch": 10.9,
"learning_rate": 5.9394999999999996e-05,
"loss": 1.6053,
"step": 1700
},
{
"epoch": 11.54,
"learning_rate": 6.289499999999999e-05,
"loss": 1.5242,
"step": 1800
},
{
"epoch": 12.18,
"learning_rate": 6.639499999999999e-05,
"loss": 1.4635,
"step": 1900
},
{
"epoch": 12.82,
"learning_rate": 6.9895e-05,
"loss": 1.4235,
"step": 2000
},
{
"epoch": 12.82,
"eval_loss": 0.4475284516811371,
"eval_runtime": 99.8092,
"eval_samples_per_second": 20.88,
"eval_steps_per_second": 20.88,
"eval_wer": 0.455098317453882,
"step": 2000
},
{
"epoch": 13.46,
"learning_rate": 6.950073529411765e-05,
"loss": 1.3615,
"step": 2100
},
{
"epoch": 14.1,
"learning_rate": 6.89860294117647e-05,
"loss": 1.2973,
"step": 2200
},
{
"epoch": 14.74,
"learning_rate": 6.847132352941176e-05,
"loss": 1.261,
"step": 2300
},
{
"epoch": 15.38,
"learning_rate": 6.795661764705882e-05,
"loss": 1.2356,
"step": 2400
},
{
"epoch": 16.03,
"learning_rate": 6.744191176470588e-05,
"loss": 1.1895,
"step": 2500
},
{
"epoch": 16.67,
"learning_rate": 6.692720588235293e-05,
"loss": 1.144,
"step": 2600
},
{
"epoch": 17.31,
"learning_rate": 6.64125e-05,
"loss": 1.1081,
"step": 2700
},
{
"epoch": 17.95,
"learning_rate": 6.589779411764706e-05,
"loss": 1.1043,
"step": 2800
},
{
"epoch": 18.59,
"learning_rate": 6.538308823529412e-05,
"loss": 1.074,
"step": 2900
},
{
"epoch": 19.23,
"learning_rate": 6.486838235294117e-05,
"loss": 1.0529,
"step": 3000
},
{
"epoch": 19.87,
"learning_rate": 6.435367647058823e-05,
"loss": 1.0452,
"step": 3100
},
{
"epoch": 20.51,
"learning_rate": 6.383897058823529e-05,
"loss": 1.0405,
"step": 3200
},
{
"epoch": 21.15,
"learning_rate": 6.332426470588235e-05,
"loss": 1.0131,
"step": 3300
},
{
"epoch": 21.79,
"learning_rate": 6.28095588235294e-05,
"loss": 1.0007,
"step": 3400
},
{
"epoch": 22.44,
"learning_rate": 6.229485294117647e-05,
"loss": 0.9744,
"step": 3500
},
{
"epoch": 23.08,
"learning_rate": 6.178014705882353e-05,
"loss": 0.9773,
"step": 3600
},
{
"epoch": 23.72,
"learning_rate": 6.126544117647059e-05,
"loss": 0.9767,
"step": 3700
},
{
"epoch": 24.36,
"learning_rate": 6.0750735294117646e-05,
"loss": 0.9495,
"step": 3800
},
{
"epoch": 25.0,
"learning_rate": 6.024117647058823e-05,
"loss": 0.9667,
"step": 3900
},
{
"epoch": 25.64,
"learning_rate": 5.9726470588235294e-05,
"loss": 0.9383,
"step": 4000
},
{
"epoch": 25.64,
"eval_loss": 0.223493292927742,
"eval_runtime": 99.8226,
"eval_samples_per_second": 20.877,
"eval_steps_per_second": 20.877,
"eval_wer": 0.23281978512061627,
"step": 4000
},
{
"epoch": 26.28,
"learning_rate": 5.921176470588234e-05,
"loss": 0.9534,
"step": 4100
},
{
"epoch": 26.92,
"learning_rate": 5.8697058823529405e-05,
"loss": 0.9294,
"step": 4200
},
{
"epoch": 27.56,
"learning_rate": 5.818235294117647e-05,
"loss": 0.9348,
"step": 4300
},
{
"epoch": 28.21,
"learning_rate": 5.766764705882352e-05,
"loss": 0.9209,
"step": 4400
},
{
"epoch": 28.85,
"learning_rate": 5.7152941176470585e-05,
"loss": 0.9104,
"step": 4500
},
{
"epoch": 29.49,
"learning_rate": 5.663823529411764e-05,
"loss": 0.9009,
"step": 4600
},
{
"epoch": 30.13,
"learning_rate": 5.61235294117647e-05,
"loss": 0.9109,
"step": 4700
},
{
"epoch": 30.77,
"learning_rate": 5.560882352941176e-05,
"loss": 0.8973,
"step": 4800
},
{
"epoch": 31.41,
"learning_rate": 5.509411764705882e-05,
"loss": 0.864,
"step": 4900
},
{
"epoch": 32.05,
"learning_rate": 5.4579411764705876e-05,
"loss": 0.8654,
"step": 5000
},
{
"epoch": 32.69,
"learning_rate": 5.406470588235294e-05,
"loss": 0.8523,
"step": 5100
},
{
"epoch": 33.33,
"learning_rate": 5.3549999999999994e-05,
"loss": 0.8794,
"step": 5200
},
{
"epoch": 33.97,
"learning_rate": 5.3035294117647056e-05,
"loss": 0.8577,
"step": 5300
},
{
"epoch": 34.62,
"learning_rate": 5.252058823529411e-05,
"loss": 0.8585,
"step": 5400
},
{
"epoch": 35.26,
"learning_rate": 5.2005882352941174e-05,
"loss": 0.8473,
"step": 5500
},
{
"epoch": 35.9,
"learning_rate": 5.149117647058823e-05,
"loss": 0.8465,
"step": 5600
},
{
"epoch": 36.54,
"learning_rate": 5.097647058823529e-05,
"loss": 0.838,
"step": 5700
},
{
"epoch": 37.18,
"learning_rate": 5.046176470588235e-05,
"loss": 0.8402,
"step": 5800
},
{
"epoch": 37.82,
"learning_rate": 4.99470588235294e-05,
"loss": 0.8432,
"step": 5900
},
{
"epoch": 38.46,
"learning_rate": 4.9432352941176465e-05,
"loss": 0.8359,
"step": 6000
},
{
"epoch": 38.46,
"eval_loss": 0.20040307939052582,
"eval_runtime": 101.1606,
"eval_samples_per_second": 20.601,
"eval_steps_per_second": 20.601,
"eval_wer": 0.20981147374822623,
"step": 6000
},
{
"epoch": 39.1,
"learning_rate": 4.891764705882353e-05,
"loss": 0.8456,
"step": 6100
},
{
"epoch": 39.74,
"learning_rate": 4.840294117647059e-05,
"loss": 0.8285,
"step": 6200
},
{
"epoch": 40.38,
"learning_rate": 4.788823529411764e-05,
"loss": 0.8054,
"step": 6300
},
{
"epoch": 41.03,
"learning_rate": 4.73735294117647e-05,
"loss": 0.82,
"step": 6400
},
{
"epoch": 41.67,
"learning_rate": 4.685882352941176e-05,
"loss": 0.8082,
"step": 6500
},
{
"epoch": 42.31,
"learning_rate": 4.6344117647058825e-05,
"loss": 0.8049,
"step": 6600
},
{
"epoch": 42.95,
"learning_rate": 4.5829411764705873e-05,
"loss": 0.8088,
"step": 6700
},
{
"epoch": 43.59,
"learning_rate": 4.5314705882352936e-05,
"loss": 0.8092,
"step": 6800
},
{
"epoch": 44.23,
"learning_rate": 4.48e-05,
"loss": 0.8212,
"step": 6900
},
{
"epoch": 44.87,
"learning_rate": 4.428529411764706e-05,
"loss": 0.7847,
"step": 7000
},
{
"epoch": 45.51,
"learning_rate": 4.3775735294117646e-05,
"loss": 0.7846,
"step": 7100
},
{
"epoch": 46.15,
"learning_rate": 4.32610294117647e-05,
"loss": 0.7966,
"step": 7200
},
{
"epoch": 46.79,
"learning_rate": 4.2746323529411764e-05,
"loss": 0.7811,
"step": 7300
},
{
"epoch": 47.44,
"learning_rate": 4.223161764705882e-05,
"loss": 0.7706,
"step": 7400
},
{
"epoch": 48.08,
"learning_rate": 4.1716911764705875e-05,
"loss": 0.7958,
"step": 7500
},
{
"epoch": 48.72,
"learning_rate": 4.120220588235294e-05,
"loss": 0.7617,
"step": 7600
},
{
"epoch": 49.36,
"learning_rate": 4.06875e-05,
"loss": 0.7662,
"step": 7700
},
{
"epoch": 50.0,
"learning_rate": 4.0172794117647055e-05,
"loss": 0.7545,
"step": 7800
},
{
"epoch": 50.64,
"learning_rate": 3.965808823529411e-05,
"loss": 0.7482,
"step": 7900
},
{
"epoch": 51.28,
"learning_rate": 3.914338235294117e-05,
"loss": 0.7633,
"step": 8000
},
{
"epoch": 51.28,
"eval_loss": 0.19603605568408966,
"eval_runtime": 98.7836,
"eval_samples_per_second": 21.097,
"eval_steps_per_second": 21.097,
"eval_wer": 0.1882221771741334,
"step": 8000
},
{
"epoch": 51.92,
"learning_rate": 3.8628676470588235e-05,
"loss": 0.7522,
"step": 8100
},
{
"epoch": 52.56,
"learning_rate": 3.811397058823529e-05,
"loss": 0.7411,
"step": 8200
},
{
"epoch": 53.21,
"learning_rate": 3.7599264705882346e-05,
"loss": 0.7443,
"step": 8300
},
{
"epoch": 53.85,
"learning_rate": 3.708455882352941e-05,
"loss": 0.7583,
"step": 8400
},
{
"epoch": 54.49,
"learning_rate": 3.656985294117647e-05,
"loss": 0.7254,
"step": 8500
},
{
"epoch": 55.13,
"learning_rate": 3.605514705882352e-05,
"loss": 0.7329,
"step": 8600
},
{
"epoch": 55.77,
"learning_rate": 3.554044117647058e-05,
"loss": 0.7216,
"step": 8700
},
{
"epoch": 56.41,
"learning_rate": 3.5025735294117644e-05,
"loss": 0.7322,
"step": 8800
},
{
"epoch": 57.05,
"learning_rate": 3.4511029411764706e-05,
"loss": 0.7405,
"step": 8900
},
{
"epoch": 57.69,
"learning_rate": 3.399632352941176e-05,
"loss": 0.7194,
"step": 9000
},
{
"epoch": 58.33,
"learning_rate": 3.3481617647058824e-05,
"loss": 0.7146,
"step": 9100
},
{
"epoch": 58.97,
"learning_rate": 3.297205882352941e-05,
"loss": 0.7128,
"step": 9200
},
{
"epoch": 59.62,
"learning_rate": 3.2457352941176466e-05,
"loss": 0.7186,
"step": 9300
},
{
"epoch": 60.26,
"learning_rate": 3.194264705882353e-05,
"loss": 0.7038,
"step": 9400
},
{
"epoch": 60.9,
"learning_rate": 3.142794117647058e-05,
"loss": 0.696,
"step": 9500
},
{
"epoch": 61.54,
"learning_rate": 3.0913235294117646e-05,
"loss": 0.707,
"step": 9600
},
{
"epoch": 62.18,
"learning_rate": 3.0398529411764704e-05,
"loss": 0.7113,
"step": 9700
},
{
"epoch": 62.82,
"learning_rate": 2.9883823529411763e-05,
"loss": 0.6997,
"step": 9800
},
{
"epoch": 63.46,
"learning_rate": 2.9369117647058822e-05,
"loss": 0.6815,
"step": 9900
},
{
"epoch": 64.1,
"learning_rate": 2.885441176470588e-05,
"loss": 0.7001,
"step": 10000
},
{
"epoch": 64.1,
"eval_loss": 0.19019167125225067,
"eval_runtime": 99.1402,
"eval_samples_per_second": 21.021,
"eval_steps_per_second": 21.021,
"eval_wer": 0.18092438678289074,
"step": 10000
},
{
"epoch": 64.74,
"learning_rate": 2.833970588235294e-05,
"loss": 0.7032,
"step": 10100
},
{
"epoch": 65.38,
"learning_rate": 2.7825e-05,
"loss": 0.7102,
"step": 10200
},
{
"epoch": 66.03,
"learning_rate": 2.7310294117647054e-05,
"loss": 0.6916,
"step": 10300
},
{
"epoch": 66.67,
"learning_rate": 2.6795588235294116e-05,
"loss": 0.6907,
"step": 10400
},
{
"epoch": 67.31,
"learning_rate": 2.6280882352941172e-05,
"loss": 0.6711,
"step": 10500
},
{
"epoch": 67.95,
"learning_rate": 2.5766176470588234e-05,
"loss": 0.6843,
"step": 10600
},
{
"epoch": 68.59,
"learning_rate": 2.525147058823529e-05,
"loss": 0.6847,
"step": 10700
},
{
"epoch": 69.23,
"learning_rate": 2.4736764705882352e-05,
"loss": 0.6533,
"step": 10800
},
{
"epoch": 69.87,
"learning_rate": 2.4222058823529407e-05,
"loss": 0.6883,
"step": 10900
},
{
"epoch": 70.51,
"learning_rate": 2.3712499999999997e-05,
"loss": 0.6779,
"step": 11000
},
{
"epoch": 71.15,
"learning_rate": 2.319779411764706e-05,
"loss": 0.6739,
"step": 11100
},
{
"epoch": 71.79,
"learning_rate": 2.2683088235294115e-05,
"loss": 0.6662,
"step": 11200
},
{
"epoch": 72.44,
"learning_rate": 2.2168382352941177e-05,
"loss": 0.6539,
"step": 11300
},
{
"epoch": 73.08,
"learning_rate": 2.1653676470588232e-05,
"loss": 0.6589,
"step": 11400
},
{
"epoch": 73.72,
"learning_rate": 2.1138970588235295e-05,
"loss": 0.6545,
"step": 11500
},
{
"epoch": 74.36,
"learning_rate": 2.062426470588235e-05,
"loss": 0.6585,
"step": 11600
},
{
"epoch": 75.0,
"learning_rate": 2.0109558823529412e-05,
"loss": 0.6472,
"step": 11700
},
{
"epoch": 75.64,
"learning_rate": 1.9594852941176468e-05,
"loss": 0.6657,
"step": 11800
},
{
"epoch": 76.28,
"learning_rate": 1.9080147058823527e-05,
"loss": 0.65,
"step": 11900
},
{
"epoch": 76.92,
"learning_rate": 1.8565441176470586e-05,
"loss": 0.652,
"step": 12000
},
{
"epoch": 76.92,
"eval_loss": 0.197893887758255,
"eval_runtime": 98.6246,
"eval_samples_per_second": 21.131,
"eval_steps_per_second": 21.131,
"eval_wer": 0.17747820798702615,
"step": 12000
},
{
"epoch": 77.56,
"learning_rate": 1.8050735294117645e-05,
"loss": 0.6559,
"step": 12100
},
{
"epoch": 78.21,
"learning_rate": 1.7536029411764703e-05,
"loss": 0.6282,
"step": 12200
},
{
"epoch": 78.85,
"learning_rate": 1.7021323529411762e-05,
"loss": 0.6251,
"step": 12300
},
{
"epoch": 79.49,
"learning_rate": 1.650661764705882e-05,
"loss": 0.6147,
"step": 12400
},
{
"epoch": 80.13,
"learning_rate": 1.599191176470588e-05,
"loss": 0.6412,
"step": 12500
},
{
"epoch": 80.77,
"learning_rate": 1.547720588235294e-05,
"loss": 0.6309,
"step": 12600
},
{
"epoch": 81.41,
"learning_rate": 1.4962499999999998e-05,
"loss": 0.6197,
"step": 12700
},
{
"epoch": 82.05,
"learning_rate": 1.4447794117647057e-05,
"loss": 0.6238,
"step": 12800
},
{
"epoch": 82.69,
"learning_rate": 1.3933088235294116e-05,
"loss": 0.635,
"step": 12900
},
{
"epoch": 83.33,
"learning_rate": 1.3418382352941174e-05,
"loss": 0.6294,
"step": 13000
},
{
"epoch": 83.97,
"learning_rate": 1.2903676470588233e-05,
"loss": 0.6171,
"step": 13100
},
{
"epoch": 84.62,
"learning_rate": 1.2388970588235292e-05,
"loss": 0.6255,
"step": 13200
},
{
"epoch": 85.26,
"learning_rate": 1.1874264705882353e-05,
"loss": 0.6238,
"step": 13300
},
{
"epoch": 85.9,
"learning_rate": 1.1359558823529412e-05,
"loss": 0.5923,
"step": 13400
},
{
"epoch": 86.54,
"learning_rate": 1.084485294117647e-05,
"loss": 0.6066,
"step": 13500
},
{
"epoch": 87.18,
"learning_rate": 1.033014705882353e-05,
"loss": 0.6143,
"step": 13600
},
{
"epoch": 87.82,
"learning_rate": 9.815441176470588e-06,
"loss": 0.6146,
"step": 13700
},
{
"epoch": 88.46,
"learning_rate": 9.300735294117647e-06,
"loss": 0.6102,
"step": 13800
},
{
"epoch": 89.1,
"learning_rate": 8.786029411764706e-06,
"loss": 0.6112,
"step": 13900
},
{
"epoch": 89.74,
"learning_rate": 8.271323529411765e-06,
"loss": 0.6025,
"step": 14000
},
{
"epoch": 89.74,
"eval_loss": 0.18655425310134888,
"eval_runtime": 99.4533,
"eval_samples_per_second": 20.955,
"eval_steps_per_second": 20.955,
"eval_wer": 0.16957226839651326,
"step": 14000
},
{
"epoch": 90.38,
"learning_rate": 7.756617647058824e-06,
"loss": 0.6147,
"step": 14100
},
{
"epoch": 91.03,
"learning_rate": 7.241911764705882e-06,
"loss": 0.5954,
"step": 14200
},
{
"epoch": 91.67,
"learning_rate": 6.7272058823529405e-06,
"loss": 0.5998,
"step": 14300
},
{
"epoch": 92.31,
"learning_rate": 6.212499999999999e-06,
"loss": 0.6078,
"step": 14400
},
{
"epoch": 92.95,
"learning_rate": 5.697794117647058e-06,
"loss": 0.6068,
"step": 14500
},
{
"epoch": 93.59,
"learning_rate": 5.183088235294118e-06,
"loss": 0.6015,
"step": 14600
},
{
"epoch": 94.23,
"learning_rate": 4.668382352941176e-06,
"loss": 0.6013,
"step": 14700
},
{
"epoch": 94.87,
"learning_rate": 4.153676470588235e-06,
"loss": 0.5948,
"step": 14800
},
{
"epoch": 95.51,
"learning_rate": 3.6389705882352937e-06,
"loss": 0.5909,
"step": 14900
},
{
"epoch": 96.15,
"learning_rate": 3.1242647058823526e-06,
"loss": 0.6129,
"step": 15000
},
{
"epoch": 96.79,
"learning_rate": 2.614705882352941e-06,
"loss": 0.5795,
"step": 15100
},
{
"epoch": 97.44,
"learning_rate": 2.1e-06,
"loss": 0.5967,
"step": 15200
},
{
"epoch": 98.08,
"learning_rate": 1.5852941176470587e-06,
"loss": 0.6013,
"step": 15300
},
{
"epoch": 98.72,
"learning_rate": 1.0705882352941175e-06,
"loss": 0.5981,
"step": 15400
},
{
"epoch": 99.36,
"learning_rate": 5.558823529411764e-07,
"loss": 0.6133,
"step": 15500
},
{
"epoch": 100.0,
"learning_rate": 4.117647058823529e-08,
"loss": 0.5932,
"step": 15600
},
{
"epoch": 100.0,
"step": 15600,
"total_flos": 5.043365508466233e+19,
"train_loss": 1.1756210368718856,
"train_runtime": 18507.9964,
"train_samples_per_second": 26.815,
"train_steps_per_second": 0.843
}
],
"max_steps": 15600,
"num_train_epochs": 100,
"total_flos": 5.043365508466233e+19,
"trial_name": null,
"trial_params": null
}